XRF Data for "Mobile carvers and mobile monuments?: a geochemical study of early medieval cross-slabs at Clonmacnoise and outlying sites, Ireland "

These XRF data are part of a geochemical study of early medieval cross slabs from Ireland. Clonmacnoise has one of the largest collections of early medieval (c. 400-1200 CE) carved stone monuments in northwest Europe. The presence of Clonmacnoise-style cross-slabs at outlying church sites within and beyond the immediate hinterland of the site provides an unmatched opportunity to investigate how stone-carving could be used to shape relationships with subsidiaries and affiliates. Geochemical analysis of this dataset collected using p-XRF demonstrates that craftspeople sometimes travelled considerable distances to carve cross-slabs at outlying sites using local stone. It also found evidence to suggest that some Clonmacnoise-style cross-slabs were carved at or near the site itself before being transported upstream via the Shannon to islands in Lough Ree and to a site immediately beyond the lake. This provides new insights into the complexity and variety of the relationships within early ecclesiastical networks, and into the organisation of the craft of stone-carving and the mobility of carvers. X-ray fluorescence (XRF) data were acquired using Thermo Fisher Scientific™ Niton™ XL2 handheld analyser with a built-in, rechargeable battery. It is equipped with multiple excitation filters to optimize sensitivity across different elemental ranges and is capable of determining a range of elements in geological materials from S to U. Specifically, the instrument's “Main Range” filter provides optimum sensitivity for elements ranging from manganese (Mn) to bismuth (Bi), while the “Low Range” filter enhances detection of lighter elements from titanium (Ti) through chromium (Cr). Although the main range filter can also detect Ti, V, and Cr, its sensitivity for these elements is lower than that of the low filter. During our measurements, the instrument operated under default filter settings, wherein it automatically alternates between filters until either the user-defined maximum analysis time is reached, or the measurement is manually terminated. The instrument uses an Ag anode X-ray tube which can operate at 45 kV and 80 uA maxima. A beam diameter of 8mm was used for our analysis. Compositional spectra were acquired on a Si-PIN detector which are processed and calibrated according to the manufacturer’s specifications using the manufacturer’s installed software. The instrument was used to analyse both cross-slabs from the archaeological sites and boulder samples from the associated sandstone lithology under the same instrumental settings. At the start of each analysis session a calibration was carried out using the Systems Check function of the detector verifying it is fully operating to specifications. For each individual analysis a minimum acquisition time of 12 seconds was used and a minimum of 12 analyses were acquired from each individual boulder and slab. These measurements were distributed across the exposed surface to account for textural and compositional heterogeneity. For the purposes of this study data were collected for the following heavier elements, Fe, Zn, Rb, Sr, Zr, Ag, Cd, Sn, Sb, and Ba. The XRF analyses were conducted over a period of four days. All samples were measured in the dry state. Although some studies have reported discrepancies in elemental concentrations between wet and dry samples (Bevins et al., 2023), these differences are typically most pronounced in the lighter elements. Since our focus is primarily on heavier elements, the influence of sample moisture is considered minimal. Nonetheless, the dry condition of the samples is noted here for transparency and reproducibility. Given the random nature of handheld XRF beam placement, some analyses may have intersected trace mineral phases that do not represent the bulk composition of the rock. As a result, a robust data-cleaning protocol was applied to minimize the impact of such outliers. The quality of each analysis was assessed using the associated 2σ (two sigma) error values generated during XRF data processing. Data points exhibiting high relative uncertainty, specifically those with error values exceeding 30%, were excluded from further analysis. Each element dataset was examined for variability, and outliers were identified using a ±2 standard deviation (2SD) threshold from the mean. In cases where the relative deviation of an analysis exceeded 20-33% for a given element, the data point was also excluded as an outlier. These criteria were applied in a flexible, element-specific manner rather than using a fixed cutoff across all elements. Despite this filtering process, the number of valid measurements per sample always remained ≥8, ensuring sufficient data for statistical analysis. Elements in the final filtered dataset included Rb, Sr, Zr, Sn, Sb, and Ba. The cleaned dataset was subsequently used to construct bivariate plots based on elemental ratios, aiding in the geochemical comparison of samples. It was decided in this study to focus on four elements, Rb, Sr, Ba and Zr, because they consistently yielded robust XRF results across most samples and are comparatively stable in sandstones. In clastic sedimentary rocks, Rb, Sr and Ba is present mainly in K-feldspar, mica and clay minerals. Zr in clastic sedimentary rocks is directly related to the presence of detrital heavy minerals, principally zircon and sphene. All of these mineral phases are present in the feldspathic arenites that typically make up ORS and Upper Carboniferous sandstone lithologies. [ur1]https://doi.org/10.1016/j.jasrep.2023.103973 [TO2]Add all refs to list at end